MRI in Practice

Parameters and trade-offs Chapter 4

Table 4.1 The results of optimizing image quality.

To optimize image Adjusted parameter Consequence
↑ scan time
Maximize SNR ↑ NEX ↓ scan time (p/matrix)
↓ resolution
↓ matrix ↓ resolution 135
↑ minimum TE
– ↑ chemical shift
↓ resolution
↑ slice thickness ↓ T1 weighting
↑ number of slices
↓ receive bandwidth ↓ T2 weighting
↓ SNR
– ↓ SNR
↑ scan time (p/matrix)
↑ FOV ↓ SNR
↑ T1 weighting
↑ TR ↓ SNR
↓ number of slices
– ↓ resolution
↑ SNR
↓ TE ↓ SNR
↑ movement artefact
Maximize resolution ↓ slice thickness ↓ SNR
(assuming a square ↑ matrix
FOV) –

↓ FOV

Minimize scan time ↓ TR

–

–

↓ phase matrix

–

↓ NEX

–

↓ slice number in volume imaging

It is very important to keep the scan time as short as possible. Again, there is no point having an
image with great SNR and spatial resolution if it took so long to acquire that the patient has moved
during the scan. Remember, any patient can move – not just a restless one. The longer the patient
is expected to lie on the table, the more likely it is that they will move.

As each system varies considerably, the following are only guidelines. The parameters given are
not etched in stone but are only meant as indicators and are appropriate at most common clinical
field strengths, i.e. 0.5 T to 1.5 T. It is inadvisable to select:

Chapter 4 MRI in Practice

Table 4.2 Parameters and their associated trade-offs. Limitation
↑ scan time
Parameter Benefit ↓ T1 weighting
↓ SNR
TR ↑ ↑ SNR ↓ number of slices
↑ number of slices ↓ SNR
136 ↓ T2 weighting
↑ scan time
TR ↓ ↓ scan time
↓ SNR
↑ T1 weighting ↓ signal averaging
↓ resolution
TE ↑ ↑ T2 weighting ↑ partial voluming
↓ SNR
TE ↓ ↑ SNR ↓ coverage
↓ resolution
NEX ↑ ↑ SNR
↓ SNR
↑ signal averaging ↓ coverage
↑ aliasing (pFOV)
NEX ↓ ↓ scan time ↑ scan time
↓ SNR if pixel small
Slice thickness ↑ ↑ SNR ↓ resolution
Slice thickness ↓ ↑ coverage
FOV ↑ ↑ resolution ↓ SNR
↓ partial voluming
FOV ↓ ↑ SNR
↑ coverage
↓ aliasing (pFOV)
↑ resolution

(p)Matrix ↑ ↑ resolution
(p)Matrix ↓
Receive bandwidth ↑ ↓ scan time
↑ SNR if pixel large
↓ chemical shift
↓ minimum TE

Parameters and trade-offs Chapter 4
Table 4.2 Continued
Benefit Limitation 137
Parameter ↑ SNR ↑ chemical shift
Receive bandwidth ↓ ↑ minimum TE
↑ area of received signal ↓ SNR
Large coil sensitive to artefacts
aliasing with small FOV
Small coil ↑ SNR ↓ area of received signal

less sensitive to artefacts

less prone to aliasing
with small FOV

• a very short TR in spin echo sequences (choose 400 ms not 200 ms)
• a very long TE (choose 100 ms not 200 ms)
• very low flip angles (choose 20° not 5°)
• very thin slices (choose 4 mm not 3 mm)
• a very small FOV (choose 120 mm not 80 mm), unless you are using a good local coil.

In most centers, the protocols selected work well and the radiologists are happy with the param-
eters set. However, it is worth remembering that, for example, a 1 mm difference in slice thickness
can make all the difference in improving SNR without noticeably reducing the spatial resolution.
Also remember that as the FOV size decreases, the dimensions of the pixel along both axes are
reduced (assuming that the system operates with a square FOV). Under these circumstances, the
FOV is the most potent controller of SNR. Using a 160 mm FOV instead of an 80 mm FOV can be
important in maintaining SNR.

If the area under examination has inherently good signal (for example the brain), and the
correct coil has been selected, it is usually possible for a fine matrix and fewer NEX to be used to
achieve good quality images in terms of SNR and spatial resolution. However, when examining an
area with inherently low signal (for example the lungs), selection of more NEX and a coarser matrix
may be necessary. Try to do all this and keep the scan time as low as possible. It is usually not
practical to have sequences that last 30 minutes each.

Volume imaging

Volume imaging is advantageous in that very small lesions can be demonstrated because the slice
thickness can be drastically reduced compared with conventional imaging, and there is no slice
gap. In conventional imaging the slice thickness affects the SNR. In volume imaging the entire
volume of tissue is excited and the volume contains no gap, the SNR is superior and so fewer NEX
can be used. The other main advantage of volumes is that as data are collected from a slab, the
slab can be manipulated to look at the anatomy within the volume in any plane and at any angle
of obliquity.

Chapter 4 MRI in Practice

138

Figure 4.36 Encoding in a volume acquisition.

The disadvantages of volume imaging are that in general, the scan times associated with them
are relatively long. For this reason, they are usually used in conjunction with faster pulse sequences.
In volume imaging, slices are sectioned out by a technique known as slice encoding (Figure 4.36).
This is another series of phase encoding steps along the slice select axis. Therefore, just as the
number of phase encoding steps increases the scan time in conventional spin echo, the number
of slices also affects the scan time in volume imaging. Therefore:

scan time = TR × NEX × number of phase encodings × number of slice encodings.

The greater the number of slices prescribed, the longer the scan time. However, this is offset
somewhat by the fact that the greater the slice number the greater the SNR, and so the NEX can
be reduced.

Volume imaging and resolution

To obtain equal resolution in every plane and at every angle of obliquity, each voxel should be
symmetrical (isotropic). That is to say, the voxel should have equal dimensions in every plane. If

Parameters and trade-offs Chapter 4

this is not true, the volume has poorer resolution in the planes other than the one in which it 139
was acquired. For example, if a FOV of 240 mm and matrix of 256 × 256 is used, each pixel has a
dimension of 0.9 mm (FOV/matrix). If the slice thickness selected is 3 mm, resolution is worse
when the voxel is viewed from the side. Under these conditions, the voxel is anisotropic.

Sometimes volumes are acquired purely because the slices are contiguous and not because
they are to be viewed in another plane, for example coronal volumes of the brain can be very
useful in detecting small temporal lobe lesions. However, they are not generally used to look at
the brain axially or coronally. In this instance, 3 mm slices at 64 locations will cover the head
adequately. In volume imaging of a joint, on the other hand, reformatting in other planes may be
paramount. Under these circumstances it is important to obtain isotropic voxels, so thinner slices
(1 mm or less) are required, although the number of slice locations may have to be increased to
cover the anatomy.

The uses of volume imaging

Volume imaging has many potential applications, but it is widely used for imaging of joints, espe-
cially the knee, where anatomy is often confusing and not strictly in plane. Volumes can be very
useful for following ligaments or other structures that cross over the imaging plane. Volumes
should also be used when looking for very small lesions. The slice thickness can be lowered to
less than 1 mm in most systems, and so extremely good resolution can be achieved. Lesions in
the temporal lobes or posterior fossa especially lend themselves to volume imaging.

Summary

• Volume imaging allows reformatting in any plane
• Isotropic voxels give equal resolution in every plane
• The scan time depends on the slice number and the TR, phase encoding number and the

NEX

• Increasing the slice number increases the SNR, but also increases the scan time
• Volume imaging increases the SNR as a whole volume of tissue is excited

Manipulating SNR, image contrast, spatial resolution and scan time is a real art and takes some
time and experience. Even after many years the operator will probably get things wrong occasion-
ally. However, perseverance is important, and eventually results in good image quality.

For questions and answers on this topic please visit the supporting
companion website for this book: www.wiley.com/go/
mriinpractice

As image quality factors and trade-offs have been explored, it is now important to understand
pulse sequences and their individual uses. These are discussed in Chapter 5.

5

Pulse sequences

Introduction 140 Conventional gradient echo 164

Spin echo pulse sequences 141 The steady state and echo

Conventional spin echo 141 formation 166

Fast or turbo spin echo 143 Coherent gradient echo 169

Inversion recovery 151 Incoherent gradient echo (spoiled) 172

Fast inversion recovery 157 Steady state free precession 175
(SSFP)

STIR (short tau inversion recovery) 157 Balanced gradient echo 179

FLAIR (fluid attenuated inversion 159 Fast gradient echo 185
recovery)

IR prep sequences 163 Single shot imaging techniques 186

Gradient echo pulse sequences 164 Parallel imaging techniques 193

Introduction

Understanding pulse sequences forms an integral part of learning MRI. Pulse sequences enable
us to control the way in which the system applies pulses and gradients. In this way, image weight-
ing and quality is determined. There are many different pulse sequences available, and each is
designed for a specific purpose. This chapter discusses the mechanisms, uses and parameters for
each of the common pulse sequences, and their advantages and disadvantages. Each manufac-
turer uses different acronyms to distinguish between individual pulse sequences, which can be
very confusing to the user. A table comparing the common acronyms for each of the main manu-
facturers is included (Table 5.2). This is provided as a guide only; it is not in any way meant to
compare the performance or specification of each system. The parameters given are general as
they depend on field strength. However, the parameters given should be suitable for most current
clinical field strengths.

MRI in Practice, Fourth Edition. Catherine Westbrook, Carolyn Kaut Roth, John Talbot.
© 2011 Blackwell Publishing Ltd. Published 2011 by Blackwell Publishing Ltd.

Pulse sequences Chapter 5

Learning point: what is 141
a pulse sequence?

The definition of a pulse sequence is a series of RF pulses, gradient applications and intervening
time periods. The RF pulses are applied for excitation purposes and, in the case of spin echo,
for rephasing purposes. The gradients are applied to spatially encode signal (see Chapter 3) and
to rephase and dephase spins depending on the type of pulse sequence and imaging option
selected. The intervening time periods refer to the time intervals between these various func-
tions, some of which are extrinsic contrast parameters that are selected at the console (see
Chapter 2). Therefore a pulse sequence is a carefully co-ordinated and timed sequence of
events to generate a particular type of image contrast. They can be thought of like dances. All
dances involve movement of the feet as a series of steps, just as all pulse sequences involve
RF pulses and gradients. However, just as the timing and the co-ordination of steps determines
the type of dance, e.g. tango, foxtrot, etc., so the timing and co-ordination of elements within
a pulse sequence determines the resultant image contrast.

Pulse sequences can generally be categorized as follows.

Spin echo pulse sequences (spins are rephased by a 180° rephasing pulse):

• conventional spin echo
• fast or turbo spin echo
• inversion recovery.

Gradient echo pulse sequences (spins are rephased by a gradient):

• coherent gradient echo
• incoherent gradient echo
• steady state free precession
• balanced gradient echo
• fast gradient echo
• echo planar imaging.

SPIN ECHO PULSE SEQUENCES
Conventional spin echo

Mechanism

This pulse sequence has previously been discussed in Chapter 2. To recap, spin echo uses a
90° excitation pulse followed by one or more 180° rephasing pulses to generate a spin echo.

Chapter 5 MRI in Practice

If only one echo is generated, a T1 weighted image can be obtained using a short TE and a
short TR. For proton density and T2 weighting, two RF rephasing pulses, generating two spin
echoes, are applied. The first echo has a short TE and a long TR to achieve proton density weight-
ing, and the second has a long TE and a long TR to achieve T2 weighting (see Figures 2.23, 2.24
and 2.25).

142

Uses

Spin echo pulse sequences are the gold standard for most imaging. They may be used for almost
every examination. T1 weighted images are useful for demonstrating anatomy because they have
a high SNR. In conjunction with contrast enhancement, however, they can show pathology. T2
weighted images also demonstrate pathology. Tissues that are diseased are generally more ede-
matous and/or vascular. They have increased water content and consequently have a high signal
on T2 weighted images and can therefore be easily identified (see Figures 2.23 to 2.26).

Parameters

T1 weighting

• Short TE 10–30 ms
• Short TR 300–700 ms
• Typical scan time 4–6 min

Proton density/T2 weighting

• Short TE 20 ms/long TE 80 ms+
• Long TR 2000 ms+
• Typical scan time 7–15 min

Advantages

• good image quality
• very versatile
• what you set is what you get (i.e. the contrast is truly based on the T1 and T2 relaxation

times of tissues)

• true T2 weighting sensitive to pathology

Disadvantages

• scan times relatively long

Pulse sequences Chapter 5

Fast or turbo spin echo

Mechanism

As the name suggests, fast or turbo spin echo is a spin echo pulse sequence, but with scan times 143
that are much shorter than conventional spin echo. To understand how fast spin echo achieves
this, it is important to recap on data acquisition in conventional spin echo (see Chapter 3). A 90°
excitation pulse is followed by a 180° rephasing pulse. Only one phase encoding step is applied
per TR on each slice and therefore only one line of K space is filled per TR (Figure 5.1).

As the scan time is a function of the TR, NEX and number of phase encodings, to reduce the
scan time one or more of these factors should be reduced. Decreasing the TR and the NEX affects
image weighting and SNR, which is undesirable. Reducing the number of phase encodings reduces
the spatial resolution, which is also a disadvantage (see Chapter 4). In fast spin echo, the scan
time is reduced by performing more than one phase encoding step and subsequently filling more
than one line of K space per TR. This is achieved by using several 180° rephasing pulses to produce
a train of echoes or echo train (Figure 5.2). At each rephasing, an echo is produced and a different
phase encoding step is performed.

In conventional spin echo, raw image data from each echo are stored in K space, and the
number of 180° rephasing pulses applied corresponds to the number of echoes produced per TR.
Each echo is used to produce a separate image (usually proton density and T2). In fast spin echo,

Figure 5.1 Spatial encoding in conventional spin echo.

Figure 5.2 The echo train.

Chapter 5 MRI in Practice

data from each echo are placed into one image. The number of 180° rephasing pulses performed
per TR corresponds to the number of echoes produced and the number of lines of K space filled.
This number is called the turbo factor or the echo train length. The higher the turbo factor, the
shorter the scan time as more phase encoding steps are performed per TR.

For example:

144 • In conventional spin echo, 256 phase matrix selected, 256 phase encodings must be applied.
Assuming 1 NEX has been selected: 256 TR times elapse to complete the scan.
• In fast spin echo, using the same parameters but selecting a turbo factor of 16, 16 phase
encoding steps are performed every TR. Therefore 256 ÷ 16 (16) TR times elapse to complete
the scan. The scan time is therefore reduced to 1/16 of the original.

At each 180°/phase encoding combination, a different amplitude of phase encoding gradient slope
is applied to fill out a different line of K space. In conventional spin echo only one line is filled per
TR, while in fast spin echo several lines corresponding to the turbo factor are filled (Figure 5.2).
Therefore K space is filled more rapidly and the scan time is reduced.

Learning point: the chest of drawers
and fast spin echo

Using the chest of drawers analogy from Chapter 3, in conventional spin echo
one drawer is opened per TR to fill one line of K space with data points. In fast
spin echo, to decrease the scan time but maintain resolution, all the drawers
must be filled (phase resolution) but more than one drawer must be opened per
TR to fill K space more quickly, reducing the scan time. This is achieved by per-
forming more than one application of the phase encoding gradient per TR, each
one to a different slope to open a different drawer.

For example, if 10 drawers are to be opened per TR, then the phase encoding
gradient must be applied 10 different times to 10 different amplitudes per TR
to open 10 different drawers. Once the drawers are opened, there must be data
to put into them. This requires producing 10 echoes, one for each drawer. To do
this, 10 different 180° pulses must be applied. The number of RF pulses corre-
sponds to the number of echoes and the number of drawers opened per TR.
This is called the echo train length or turbo factor and indicates how much faster
the scan is compared with conventional spin echo, i.e. a turbo factor of 16 indi-
cates 16 drawers are opened per TR and the scan time is 16 times faster than
conventional spin echo.

Weighting in fast spin echo

The echoes are generated at different TE times and therefore data collected from them have vari-
able weighting. All these data are stored and placed into one image. So how is a fast spin echo
sequence weighted correctly? The TE selected is only an effective TE. In other words, it is the TE

Pulse sequences Chapter 5

145

Figure 5.3 Phase encoding gradient slopes.

at which the operator wishes to weight the resultant image. To achieve this weighting, the system
orders the phase encoding steps so that steep or shallow slopes are applied to the various echoes
produced. As described in Chapter 3, each phase encoding step applies a different slope of gradi-
ent to phase shift the signal by a different amount. If 256 phase encodings are performed, the
phase encoding gradient is switched on to varying degrees from +128 to −128 (or +128 to −127
if the 0 line is included) (Figure 5.3).

Very steep phase encoding slopes reduce the amplitude of the resultant echo. Shallow phase
encoding slopes result in an echo that has maximum signal amplitude (Figure 5.4) (see Chapter
3). The system orders the phase encodings so that the shallow slopes that produce maximum
signal are centered on the effective TE selected. The steep slopes that produce much smaller
signal amplitude are placed away from the effective TE. The resultant image contains data from
all the echoes in the echo train, but data from echoes collected around the effective TE have more
impact on image contrast as they fill the central lines of K space, which produce the greatest signal
amplitude. Data from echoes collected at the wrong weighting (other TEs) have much less of an
effect on the contrast, as they fill the outer lines of K space and therefore have a smaller signal
amplitude and a greater spatial resolution (Figure 5.5).

If a TE of 100 ms is selected, with a TR 4000 ms and a turbo factor of 16, T2 weighting is required.
The shallowest phase encodings are performed on echoes occurring around 100 ms. Data acquired
from these phase encodings have a TE at or close to 100 ms. Phase encodings performed at the
very beginning and end of the echo train are steep, and the signal amplitude of these echoes is
small. They contain either proton density or very heavily T2 weighted data, which are present in
the image but whose impact is less predominant.

Uses

Generally speaking, the contrast seen in fast spin echo images is similar to spin echo and, there-
fore, these sequences are useful in most clinical applications. In the central nervous system, pelvis

Chapter 5 MRI in Practice

146

Figure 5.4 Phase encoding vs signal amplitude.

and musculoskeletal regions, fast spin echo has largely replaced spin echo especially for T2 weight-
ing. In the chest and abdomen, respiratory artefact is sometimes troublesome if respiratory
compensation techniques are not compatible with fast spin echo software. This is offset by the
fact that the shorter scan times of fast spin echo enable images to be produced while the patient
holds their breath.

There are, however, two contrast differences between spin echo and fast spin echo, both of
which are due to the repeated, closely spaced 180° pulses of the echo train. First, fat remains
bright on T2 weighted images due to the multiple RF pulses, which reduce the effects of spin–spin
interactions in fat (J coupling) (Figure 5.6). However, fat saturation techniques can be used to
compensate for this (see Chapter 6). Second, the repeated 180° pulses can increase magnetization
transfer effects so that muscle, for example, appears darker on fast spin echo images than in
conventional spin echo. In addition, the multiple 180° pulses reduce magnetic susceptibility
effects, which can be detrimental when looking for small hemorrhages.

Image blurring may occur in fast spin echo images at the edges of tissues with different T2
decay values. This is because each line of K space filled during an echo train contains data from
echoes with a different TE. When using long echo trains, late echoes that have a low signal ampli-
tude contribute to the resolution of K space. If these echoes are negligible, then resolution is lost
from the image and blurring occurs. This, however, may be reduced by decreasing the spacing
between echoes and/or the turbo factor. In addition, artefact from metal implants is significantly
reduced when using fast spin echo because the repeated 180° RF pulses compensate for field
inhomogeneity (see Chapter 7).

Pulse sequences Chapter 5

147

Figure 5.5 K space filling and phase re-ordering.

Parameters

These are similar to conventional spin echo. However, the turbo factor now plays an important
role in image weighting. The higher the turbo factor, the shorter the scan time, but the result-
ant image has more of a mixture of weighting because there are more data collected at the
wrong TE. This is not as important in T2 weighted scans, as the proton density data are offset
somewhat by the heavily T2 weighted data. In T1 and proton density weighting, on the other
hand, larger turbo factors place too much T2 weighting in the image and hence shorter turbo
factors must be used. The scan time savings in T1 weighted imaging are therefore not as great
as with T2 weighting.

For T1 weighting (Figure 5.7):

• TR 300–700 ms
• effective TE minimum
• turbo factor 2–8.

Chapter 5 MRI in Practice

For PD weighting (Figure 5.8):

• TR 3000–10 000 ms (depending on required slice number)
• effective TE minimum
• turbo factor 2–8.

148 For T2 weighting (Figure 5.9):

• TR 3000–10 000 ms (depending on required slice number)
• effective TE 80–140 ms
• turbo factor 12–30.

The TR of fast spin echo is often much longer than that used in conventional spin echo. The
180° RF pulses take time to perform and so fewer slices are available for a given TR. As the
turbo factor increases, the number of slices available per TR decreases, and sometimes the TR
has to be significantly increased to achieve the required slice number. In T1 weighting, increas-
ing the TR reduces the weighting, and so in these circumstances we need to keep the TR short
and to perform several acquisitions to obtain coverage of anatomy. The longer TR associated
with fast spin echo somewhat offsets the reduction in scan time achieved, but is far less sig-
nificant than the huge scan time savings produced by long echo trains.

Summary

Short turbo factor

• decreased effective TE
• increased T1 weighting
• longer scan time
• more slices per TR
• reduced image blurring

Long turbo factor

• increased effective TE
• increased T2 weighting
• reduced scan time
• reduced slice number per TR
• increased image blurring

Advantages

• scan times greatly reduced
• high-resolution matrices and multiple NEX can be used
• image quality improved
• increased T2 information

Pulse sequences Chapter 5

Disadvantages 149

• some flow and motion affects increased
• incompatible with some imaging options
• fat bright on T2 weighted images
• image blurring with very long echo trains

Figure 5.6 Sagittal T2 weighted fast spin echo sequence through the pelvis. Note that both fat
and water have high signal intensity.

Chapter 5 MRI in Practice

150

Figure 5.7 Sagittal T1 weighted fast spin echo image of the knee.

Single shot fast spin echo (SS-FSE)

It is possible to acquire fast spin echo images in even shorter scan times by using a technique
known as single shot fast spin echo (SS-FSE). In this technique all the lines of K space are acquired
in one TR. SS-FSE combines a partial Fourier technique with fast spin echo. Half of the lines of K
space are acquired in one TR and the other half are transposed. This technique yields a reduction
in imaging time as all the image data are acquired in one TR. However, there is a SNR penalty.
Currently the highest turbo factor used in single shot imaging is 728. Another consideration when
using long echo trains is that the specific absorption rate (SAR) is significantly increased by apply-
ing so many successive 180° pulses. This usually manifests itself in a reduction in the number of
permissible slices and it can therefore be difficult to get the required coverage in a single acquisi-
tion. It is possible on most systems to reduce the size of the refocusing angle to as low as 120°.
This reduces the SAR significantly (which is proportional to the square of the flip angle) but also
reduces the SNR. However, the benefit of being able to obtain more slices per acquisition due to
a reduction in SAR may outweigh the decrease in SNR.

Pulse sequences Chapter 5

151

Figure 5.8 Sagittal PD weighted fast spin echo image of the knee.

Driven equilibrium Fourier transform

In another modification of FSE (which some manufacturers call DRIVE, RESTORE or FR-FSE), a
reverse flip angle excitation pulse is applied at the end of the echo train. This drives any transverse
magnetization into the longitudinal plane so that it is available for excitation at the beginning of
the next TR period. Therefore it is not necessary to wait long periods for T1 relaxation to occur.
Some manufacturers rephase the transverse magnetization with a 180° pulse before the restora-
tion 90° pulse is applied. As water has the longest T1 and T2 times, most of this magnetization is
composed of water and therefore this has a higher signal intensity on the resultant images. This
sequence produces an increase in signal intensity in fluid-based structures such as cerebrospinal
fluid (CSF) when using shorter TRs than normal in FSE (Figures 5.10 and 5.11).

Inversion recovery

Mechanism

Inversion recovery (IR) was developed in the early days of MRI to provide good T1 contrast on
low field systems. However, the scan times were relatively long and when high field superconduct-

Chapter 5 MRI in Practice

152

Figure 5.9 Sagittal T2 weighted fast spin echo image of the knee.

ing systems were widely used, this sequence became somewhat redundant. However, it has re-
emerged combined with fast spin echo to produce images in a few minutes. It is usually used to
suppress the signal from certain tissues in conjunction with long TEs and T2 weighting, although
at low field it is still used for T1 contrast. All varieties are discussed here.

Inversion recovery is a pulse sequence that begins with a 180° inverting pulse. This inverts the
NMV through 180° into full saturation. When the inverting pulse is removed, the NMV begins to
relax back to B0. A 90° excitation pulse is then applied at a time from the 180° inverting pulse
known as the TI (time from inversion) (Figure 5.12). The resultant FID is then rephased by a 180°
pulse to produce a spin echo at time TE (Figure 5.13).

The contrast of the resultant image depends primarily on the length of the TI. If the 90° excita-
tion pulse is applied after the NMV has relaxed back through the transverse plane, the contrast
in the image depends on the amount of longitudinal recovery of each vector (as in spin echo).
The resultant image is heavily T1 weighted because the 180° inverting pulse achieves full satura-
tion and ensures a large contrast difference between fat and water (Figure 5.14). If the 90° excita-

Pulse sequences Chapter 5

153

Figure 5.10 The DRIVE pulse sequence.

tion pulse is not applied until the NMV has reached full recovery, a proton density weighted image
results, as both fat and water have fully relaxed (Figure 5.15).

Uses

Inversion recovery was conventionally used to produce heavily T1 weighted images to demon-
strate anatomy (Figure 5.16). The 180° inverting pulse produces a large contrast difference
between fat and water because full saturation of the fat and water vectors is achieved at the
beginning of each repetition. Therefore tissues begin their recovery from full saturation as opposed
to from the transverse plane as in conventional spin echo. This allows more time for differences
in the T1 recovery times between tissues to show up, and therefore IR pulse sequences produce
heavier T1 weighting than conventional spin echo. As the use of gadolinium primarily shortens
the T1 times of certain tissues, IR pulse sequences increase the signal from structures that have
enhanced as a result of a contrast injection.

Chapter 5 MRI in Practice

Parameters

When inversion recovery is used to produce predominantly heavily T1 weighted images at low
field, the TE controls the amount of T2 decay, and so it is usually kept short to minimize T2
effects. However, it can be lengthened to give tissues with a long T2 a bright signal. This is

154 called pathology weighting and produces an image that is predominantly T1 weighted, but

where pathological processes appear bright. The TI is the most potent controller of contrast in
the inversion recovery sequence. Medium TI values give T1 weighting, but as this is lengthened
the image becomes more proton density weighted. The TR should always be long enough to
allow full recovery of the NMV before the next inverting pulse is applied. If this is not so, indi-
vidual vectors recover to different degrees, and the weighting is affected. For example, at 1T
to achieve full recovery of the NMV, the TR should be longer than 3000 ms. Most systems now
use inversion recovery fast spin echo (see below).

T1 weighting

• medium TI 400–800 ms (varies at different field strengths)
• short TE 10–20 ms
• long TR 3000 ms+
• average scan time 5–15 min.

Proton density weighting

• long TI 1800 ms
• short TE 10–20 ms
• long TR 3000 ms+
• average scan time 5–15 min.

Pathology weighting

• medium TI 400–800 ms
• long TE 70 ms+
• long TR 3000 ms+
• average scan time 5–15 min.

Advantages

• very good SNR as the TR is long
• excellent T1 contrast

Disadvantages

• long scan times unless used in conjunction with fast spin echo

155

Figure 5.11 Axial DRIVE image through the right internal auditory meatus. Note high signal
intensity in CSF.

Figure 5.12 The 180° inverting pulse in an inversion recovery sequence.

Chapter 5 MRI in Practice

156

Figure 5.13 The inversion recovery sequence.

Figure 5.14 T1 weighting in inversion recovery.

Pulse sequences Chapter 5

Refer to animation 5.1 on the supporting companion website for this
book: www.wiley.com/go/mriinpractice

157

Figure 5.15 PD weighting in inversion recovery.

Fast inversion recovery

In this sequence modification the 180° inverting pulse is followed after the TI time by the 90° excita-
tion pulse and the train of 180° RF pulses to fill out multiple lines of K space as in fast spin echo.
This greatly reduces the scan time and enabled a re-emergence of this sequence in clinical imaging.
However, instead of being used to produce T1 weighted images, fast inversion recovery is usually
used to suppress signal from certain tissues in conjunction with T2 weighting so that water and
pathology return a high signal. The two main sequences in this category are STIR and FLAIR.

STIR (short tau inversion recovery)

Mechanism

STIR is an inversion recovery pulse sequence that uses a TI (also called tau) that corresponds to
the time it takes fat to recover from full inversion to the transverse plane so that there is no longi-
tudinal magnetization corresponding to fat. This is called the null point (Figure 5.17). When the 90°

Chapter 5 MRI in Practice

158

Figure 5.16 Axial T1 weighted inversion recovery sequence through the brain. A TI of 700 ms
was used.

excitation pulse is applied, because there is no longitudinal component of fat, there is no transverse
component after excitation and the signal from fat is nulled. A TI of 100–175 ms achieves fat sup-
pression, although this value varies slightly at different field strengths. The TI required to null the
signal from a tissue is 0.69 times its T1 relaxation time. It is important to note that STIR should not
be used in conjunction with contrast enhancement, which shortens the T1 times of enhancing
tissues, making them bright. The T1 times of these structures are shortened so that they approach
the T1 time of fat. In a STIR sequence therefore, enhancing tissue may also be nulled.

Uses

STIR is an extremely important sequence in musculoskeletal imaging because normal bone, which
contains fatty marrow, is suppressed and lesions within bone such as bone bruising and tumors
are seen more clearly (Figures 5.18 and 5.19). It is also a very useful sequence for suppressing fat
in general MR imaging (see Chapter 6).

Pulse sequences Chapter 5

159

Figure 5.17 STIR.

Parameters

• Short TI (tau) 150–175 ms (to suppress fat depending on field strength)
• Long TE 50 ms+ (to enhance signal from pathology)
• Long TR 4000 ms+ (to allow full recovery)
• Long turbo factor 16–20 (to enhance signal from pathology)
• Average scan time 5–15 min

FLAIR (fluid attenuated inversion recovery)

Mechanism

FLAIR is another variation of the inversion recovery sequence. In FLAIR, selecting a TI correspond-
ing to the time of recovery of CSF from 180° to the transverse plane nulls the signal from CSF.
There is no longitudinal magnetization present in CSF. When the 90° excitation pulse is applied,
because there is no longitudinal component of CSF there is no transverse component after excita-
tion and the signal from CSF is nulled. FLAIR is used to suppress the high CSF signal in T2 weighted
images so that pathology adjacent to CSF is seen more clearly. A TI of 1700–2200 ms achieves CSF
suppression (although this varies slightly at different field strengths and is calculated by multiply-
ing the T1 relaxation time of CSF by 0.69).

Chapter 5 MRI in Practice

160

Figure 5.18 Sagittal STIR sequence of the knee. Normal bone marrow has been nulled.
Synovial fluid in the joint has a high signal as the TE is long and the image is therefore T2
weighted.

Uses

FLAIR is used in brain and spine imaging to see periventricular and cord lesions more clearly,
because high signal from CSF that lies adjacent is nulled. It is especially useful in visualizing mul-
tiple sclerosis plaques, acute sub-arachnoid hemorrhage and meningitis (Figure 5.20). Sometimes
gadolinium is given to enhance pathology. This oddity (gadolinium enhancement in T2 weighted
images) may be due to the fact that the long echo trains used in FLAIR sequences cause fat to
remain bright on T2 weighted images. As gadolinium reduces the T1 relaxation time of enhancing
tissue so that it is similar to fat, enhancing tissue may appear brighter than when gadolinium is
not given. Another modification of this sequence in brain imaging is selecting a TI time that cor-
responds to the null point of white matter. This nulls the signal from normal white matter so that
lesions within it appear much brighter by comparison. This sequence (which requires a TI of about
300 ms) is very useful for white matter lesions such as periventricular leukomalacia and for con-
genital gray/white matter abnormalities (Figure 5.21).

161

Figure 5.19 Sagittal STIR sequence of the lumbar spine using similar parameters to Figure 5.18.

Chapter 5 MRI in Practice

162

Figure 5.20 Axial FLAIR image through the brain.

Parameters

• long TI 1700–2200 ms (to suppress CSF depending on field strength)
• long TE 70 ms+ (to enhance signal from pathology)
• long TR 6000 ms+ (to allow full recovery)
• long turbo factor 16–20 (to enhance signal from pathology)
• average scan time 13–20 min

Pulse sequences Chapter 5

163

Figure 5.21 Coronal IR sequence using a TI that nulls white matter.

IR prep sequences

There are two further modifications of fast IR that were specifically developed to null blood in
cardiac imaging (see Chapter 8). Double IR prep begins with two 180° pulses. One is non-slice
selective and inverts all spins in the imaging volume, and the other is slice selective and re-inverts
spins within a slice. A TI corresponding to the null point of blood (about 800 ms) completely nulls
the signal from blood in the slice so that black blood imaging results. This is useful when looking
at the morphology of the heart and great vessels. Triple IR prep adds a further inverting pulse at
the TI of fat (about 150 ms) to null fat and blood together. This is useful when determining fatty
infiltration of the heart walls (see Figure 8.3).

Chapter 5 MRI in Practice

GRADIENT ECHO PULSE SEQUENCES

Conventional gradient echo

Mechanism

164

Gradient echo pulse sequences have been discussed in Chapter 2. To recap, gradient echo
sequences use variable flip angles so that the TR and therefore the scan time can be reduced
without producing saturation. T2* and proton density weighting, which are normally associated
with long TRs and scan times, can therefore be acquired using short TRs because the sequence
begins with a flip angle less than 90°. A gradient rather than a 180° rephasing RF pulse is used to
rephase the FID. The frequency encoding gradient is used for this purpose because it is quicker
to apply than a 180° pulse and therefore the minimum TE can be reduced. The frequency encod-
ing gradient is initially applied negatively to increase dephasing of the FID, and then its polarity
is reversed producing rephasing of the gradient echo. However, the gradient does not compensate
for magnetic field inhomogeneities, so the resultant echo displays a great deal of T2* information
(Figure 5.22).

Figure 5.22 A basic gradient echo sequence showing how a bipolar application of the
frequency encoding gradient produces a gradient echo.

Pulse sequences Chapter 5

Uses

Gradient echo pulse sequences can be used to acquire T2*, T1 and proton density weighting. 165
However, there is always some degree of T2* weighting due to the absence of a 180° rephasing pulse.
Gradient echo sequences allow for a reduction in the scan time as the TR is greatly reduced. They
can be used for single-slice or volume breath-hold acquisitions in the abdomen, and for dynamic
contrast enhancement. They are very sensitive to flow as gradient rephasing is not slice selective, so
flowing nuclei always give a signal, as long as they have been previously excited (see Chapter 6).
Because of this, gradient echo sequences may be used to produce angiographic-type images.

Parameters

The flip angle, in conjunction with the TR, determines the degree of saturation
and therefore T1 weighting. To prevent saturation (necessary for T2* and proton
density weighting) the flip angle should be small and the TR long enough to
permit full recovery (although if the flip angle is small full recovery occurs using
a much shorter TR than in spin echo imaging). If saturation and therefore T1
weighting is required, the flip angle should be large and the TR short, so that full
recovery cannot occur. The TE controls the amount of T2* dephasing. To mini-
mize T2* the TE should be short. To maximize it, the TE should be long (see heat
analogy in Chapter 2 and Figures 2.36 and 2.37).

T1 weighting

• large flip angle 70–110° (to maximize saturation)
• short TR less than 50 ms (to maximize saturation)
• short TE 1–5 ms (to minimize T2*)
• average scan time several seconds to minutes.

T2* weighting

• small flip angle 5–20° (to minimize saturation)
• long TR 200 ms+ (to minimize saturation)
• long TE 15–25 ms (to maximize T2*)
• average scan time several seconds to minutes.

Proton density weighting

• small flip angle 5–20° (to minimize saturation)
• long TR 200 ms+ (to minimize saturation)
• short TE 5–10 ms (to maximize T2*)
• average scan time several seconds to minutes.

In conventional gradient echo the TR does not always affect image contrast.
Once a certain value of TR has been exceeded, the NMV recovers fully, regard-
less of the flip angle selected. Under these circumstances the flip angle and TE
control the degree of saturation and dephasing respectively.

Chapter 5 MRI in Practice

The steady state and echo formation

The steady state is a term used in many scientific contexts. It is defined as the stable condition
that does not change over time. For example, if a pot of water is placed on a stove, the stove will
gradually heat up the pot and the water. In addition heat energy is lost from the pot and the water
through processes such and conduction, convection and evaporation. If the amount of heat energy

166 from the stove equals the amount of heat energy lost by convection, conduction and evaporation,

then the temperature of the pot and water will remain constant and stable. This is an example of
the steady state because the energy ‘in’ equals the energy ‘out’.

In MRI, energy is given to hydrogen during excitation and, as described by the classical theory,
the amount of energy applied is indicated by the flip angle. Energy is lost by hydrogen through
spin lattice energy transfer and the amount of energy lost is determined by the TR. Therefore by
selecting a certain combination of TR and flip angle, we can insure that the overall energy of
hydrogen remains constant as the energy ‘in’ as determined by the flip angle equals the energy
‘out’ as determined by the TR. There are, therefore, critical values of flip angle and TR to maintain
the steady state (Figure 5.23).

As RF has a low frequency and hence low energy, for most values of flip angle very short TRs
are required to achieve the steady state. In fact the TRs required are shorter than the T1 and T2
relaxation times of the tissues. There is therefore no time for transverse magnetization to decay
before the pulse sequence is repeated. Generally, flip angles of 30° to 45° in conjunction with a
TR less than 50 ms achieve the steady state.

Figure 5.23 The steady state.

Pulse sequences Chapter 5

Table 5.1 T1 and T2 relaxation times and signal intensity of brain tissue in the steady state
at 1 T.

Tissue T1 time (ms) T2 time (ms) T1/T2 Signal intensity

Water 2500 2500 1↑

Fat 200 100 0.5 ↑ 167

Cerebrospinal fluid 2000 300 0.15 ↓

White matter 500 100 0.2 ↓

In the steady state, there is co-existence of both longitudinal and transverse magnetization. In
particular, the transverse component of magnetization does not have time to decay during the
pulse sequence and builds up over successive TRs. This transverse magnetization is produced as
a result of previous excitations, but remains over several TR periods in the transverse plane. It is
called the residual transverse magnetization and it affects image contrast as it induces a voltage
in the receiver coil. It affects image contrast as it results in tissues with long T2 times (such as
water), appearing bright on the image. Generally speaking, as the TR is so short, magnetization
in tissues does not have time to reach its T1 or T2 times before the next excitation pulse is applied.
Therefore in the steady state image contrast is not due to differences in T1 and T2 times of tissues
but rather to the ratio of T1 to T2, i.e. in tissues where T1 and T2 times are similar, the signal
intensity is high.

In the human body, fat and water have this parity (fat, very short T1 and T2 times; water, very
long T1 and T2 times) and therefore return high signal intensity in steady state sequences (Table
5.1). Other tissues such as muscle return a lower signal intensity because they do not have a
similar T1 and T2 decay time. Most gradient echo sequences use the steady state as the shortest
TR and scan time is achieved. Gradient echo sequences are classified according to whether the
residual transverse magnetization is in phase (coherent) or out of phase (incoherent).

Learning point: echo formation

The steady state involves repeatedly applying RF pulses at time intervals less than the T2 and
T1 times of all the tissues. There is therefore a build-up of residual transverse magnetization
and in the steady state, this is rephased by RF pulses to produce a spin echo.

This happens because every RF pulse (regardless of its net amplitude as determined by
the flip angle) contains energies that are sufficient to rephase transverse magnetization (they
also contain energies that cause resonance but this is not relevant for this explanation). These
energies rephase the residual transverse magnetization left over from previous RF excita-
tion pulses to form a spin echo. This occurs at exactly the same time as the next RF pulse
because the residual transverse magnetization takes the same time to rephase as it took to
dephase in the first place. Therefore, when utilizing the steady state, the TR equals the tau of
the spin echo.

Chapter 5 MRI in Practice

Look at Figures 5.24 and 5.25. You will see from these diagrams that a train of RF pulses
generates two signals:

• a FID, which occurs as a result of the withdrawal of the previous RF pulse and, once rephased,
contains either T2* or T1 information depending on the TE

168 • a stimulated echo whose peak occurs at the same time as a subsequent RF pulse and con-

tains T2* and T2 information.

The first RF pulse (RF pulse 1, shown in red) produces a FID (also shown in red). The second RF
pulse (RF pulse 2, shown in orange) also produces a FID (also shown in orange). However,
because the TR between RF pulse 1 and 2 is shorter than the relaxation times of the tissues,
transverse magnetization is still present when the RF pulse 2 is applied. RF pulse 2 produces a
FID and rephases the residual transverse magnetization still present from the first RF pulse. A
spin or stimulated echo is therefore produced. This occurs at the same time as the third RF
pulse (RF pulse 3, shown in blue) because the time for rephasing this transverse magnetization
is the same time it took to dephase. Therefore at RF pulse 3 there are two signals: a FID (shown
in blue) produced as a result of the excitation properties of RF pulse 3, and a spin echo (shown
in red) that was produced by RF pulse 1 and rephased by RF pulse 2.

Any two RF pulses produce a spin echo. The first RF pulse excites the nuclei regardless
of its net amplitude; the second RF pulse rephases the FID and any residual magnetization
present to produce a spin echo (Figures 5.24 and 5.25). These echoes are termed Hahn or
stimulated echoes depending on the amplitude of the RF pulses involved. Any two 90° RF
pulses produce a Hahn echo (after Erwin Hahn who discovered them). Any two RF pulses with
varying amplitude, i.e. with flip angles other than 90°, are called stimulated echoes. This
type of echo is used in steady state gradient echo sequences. Most gradient echo sequences
contain data from FIDs and stimulated echoes. Their contrast is determined by which of these
are digitized and used in the resultant image. In practice, echo production is so rapid that the
tails of FID signals merge with stimulated echoes, resulting in a continuous signal of varying
amplitude. However, in the interests of simplicity, the diagrams in this chapter show them
separately.

Summary

• The steady state is created when the TR is shorter than the relaxation times of tissues

and the energy ‘in’ as determined by the flip angle equals the energy ‘out’ during the TR

period

• Residual magnetization therefore builds up in the transverse plane
• The residual transverse magnetization is rephased by subsequent RF pulses to produce

stimulated echoes

• The resultant image contrast is due to the ratio of T1 to T2 in a particular tissue and whether

the FID and or the stimulated echo are sampled

Pulse sequences Chapter 5
Figure 5.24 Echo formation in the steady state I.
169

Figure 5.25 Echo formation in the steady state II.

Coherent gradient echo

Mechanism

Coherent gradient echo pulse sequences use a variable flip angle excitation pulse followed by
gradient rephasing to produce a gradient echo. The steady state is maintained by selecting a TR
shorter than the T1 and T2 times of the tissues. There is therefore residual transverse magnetiza-
tion left over when the next excitation pulse is delivered. These sequences keep this residual
magnetization coherent by a process known as rewinding. Rewinding is achieved by reversing the
slope of the phase encoding gradient after readout (Figure 5.26). This results in the residual mag-
netization rephasing, so that it is in phase at the beginning of the next repetition.

Chapter 5 MRI in Practice

170

Figure 5.26 The coherent gradient echo sequence.

The rewinder gradient rephases all transverse magnetization regardless of when it was created.
Therefore the resultant echo contains information from the FID and the stimulated echo. These
sequences can therefore be used to achieve T1 or T2* weighted images, although traditionally
they are used in conjunction with a long TE to produce T2* weighting.

Uses

Coherent gradient echo pulse sequences usually produce rapid images that are T2* weighted
(Figures 5.27 and 5.28). As water is bright they are often said to give an angiographic, myelo-
graphic or arthrographic effect. They can be used to determine whether a vessel is patent, or
whether an area contains fluid. They can be acquired slice by slice, or in a 3D volume acquisition.
As the TR is short, slices can be acquired in a single breath hold.

Parameters

To maintain the steady state:

• flip angles 30–45°
• TR 20–50 ms.

Pulse sequences Chapter 5

To maximize T2*: 171

• long TE 15–25 ms
• use gradient moment rephasing to accentuate T2* and reduce flow artefact (see

Chapter 6)

• average scan time: seconds for single slice, minutes for volumes.

Advantages

• very fast scans, breath-holding possible
• very sensitive to flow so good for angiography
• can be acquired in a volume acquisition

Disadvantages

• reduced SNR in 2D acquisitions
• magnetic susceptibility increases (see Chapter 7)
• loud gradient noise

Figure 5.27 Axial breath-hold coherent gradient echo sequence through the abdomen showing
vessel patency in the aorta and inferior vena cava.

Chapter 5 MRI in Practice

172

Figure 5.28 Axial coherent gradient echo sequence through the cervical spine. Note the high
signal in the carotid arteries and jugular veins.

Incoherent gradient echo (spoiled)

Mechanism

Incoherent gradient echo pulse sequences begin with a variable flip angle excitation pulse and
use gradient rephasing to produce a gradient echo. The steady state is maintained so that residual
transverse magnetization is left over from previous repetitions. These sequences dephase or spoil
this magnetization so that its effect on image contrast is minimal. Only transverse magnetization
from the previous excitation is used, enabling T1 contrast to dominate. There are two ways to
achieve spoiling, as follows.

RF spoiling. In this sequence RF is transmitted at a particular frequency to excite a slice and at
a specific phase. The receiver coil digitally communicates with the transmit coil and only frequen-
cies from echoes that have just been created by the excitation pulse are digitized. Using the watch
analogy from Chapter 1, disregard the precessional rotation of transverse magnetization for the pur-
poses of this explanation and look at Figure 5.29. The first RF excitation pulse applied to a particu-
lar slice has a phase of 3 o’clock. This means that the resultant transverse magnetization is created
at 3 o’clock in the transverse plane. Spins dephase and are rephased by a gradient to produce a
gradient echo. The receiver coil, which is situated in the transverse plane, samples frequencies
within this echo and data from them are sent to K space to produce the resultant image.

Pulse sequences Chapter 5

173

Figure 5.29 RF spoiling in the incoherent gradient echo sequence.

A short TR period later the process is repeated, but this time the RF excitation pulse creates
transverse magnetization at a different phase, such as 6 o’clock. Spins dephase and are rephased
by a gradient to produce a second gradient echo. The receiver coil samples frequencies within
this echo and data from them are sent to K space to produce the resultant image. However, as
the TR was so short, magnetization created at 3 o’clock is still present as it has not had time to
decay. This is the residual transverse magnetization, but because it has a different phase to the
transverse magnetization just created, it is not sampled and therefore does not impact image
contrast. This is RF spoiling and enables only information from the most recently created mag-
netization to affect image contrast.

Gradient spoiling. Gradients can be used to dephase and rephase the residual magnetization.
Gradient spoiling is the opposite of rewinding. In gradient spoiling, the slice select, phase encoding
and frequency encoding gradients can be used to dephase the residual magnetization, so that it
is incoherent at the beginning of the next repetition. In this way, T2* or T2 effects are reduced.
Generally, the uses and parameters involved in these sequences are similar to those used in RF
spoiling. However, most manufacturers use RF spoiling in incoherent gradient echo sequences.

Chapter 5 MRI in Practice

174

Figure 5.30 Coronal incoherent gradient echo sequence through the brain. This was acquired
as part of a volume acquisition enabling T1 weighted high-resolution imaging.

Uses

As the stimulated echo that contains mainly T2* and T2 information is spoiled, RF spoiled pulse
sequences produce T1 or proton density weighted images, although fluid may have a rather high
signal due to gradient rephasing (Figure 5.30). They can be used for 2D and volume acquisitions,
and as the TR is short, 2D acquisitions can be used to acquire T1 weighted breath-hold images.
RF spoiled sequences demonstrate good T1 anatomy and pathology after gadolinium.

Parameters

To maintain the steady state:

• flip angle 30–45°
• TR 20–50 ms.

Pulse sequences Chapter 5

To maximize T1: 5–10 ms
several seconds for single slice, minutes for volumes.
• short TE
• average scan time

175

Advantages

• can be acquired in a volume or 2D
• breath holding possible
• good SNR and anatomical detail in volume
• can be used after gadolinium contrast injection

Disadvantages

• SNR poor in 2D
• loud gradient noise

Steady state free precession (SSFP)

Mechanism

In gradient echo sequences the TE is not long enough to measure the T2 time of tissues as a TE
of at least 70 ms is required for this. In addition, gradient rephasing is so inefficient that any echo
is dominated by T2* effects and therefore true T2 weighting cannot be achieved. The SSFP
sequence overcomes this problem to obtain images that have a sufficiently long TE and less T2*
than in other steady state sequences. This is achieved in the following manner.

As previously described, every RF pulse, regardless of its net magnitude, contains energies that
have sufficient magnitude to rephase spins and produce a stimulated echo. However, in SSFP we
need to digitize frequencies only from this stimulated echo and not from the FID. To do this, the
stimulated echo must be repositioned so that it does not occur at the same time as the subse-
quent excitation pulse. This is achieved by applying a rewinder gradient, which speeds up the
rephasing so that the stimulated echo occurs sooner (Figure 5.31).

The resultant echo demonstrates more true T2 weighting than conventional gradient echo
sequences. This is because:

• The TE is now longer than the TR. In SSFP, there are usually two TEs.
• The actual TE is the time between the echo and the next excitation pulse.
• The effective TE is the time from the echo to the excitation pulse that created its FID.

Therefore:

effective TE = (2 × TR) − TE.

If the TR is 50 ms and the TE is 10 ms, then:

Chapter 5 MRI in Practice

176

Figure 5.31 The SSFP sequence. Note how a rewinder gradient repositions each spin echo so
that it no longer occurs at the same time as an excitation pulse but just before it. It can therefore
be sampled on its own and the effects of the FID are eliminated.

effective TE = (2 × 50) − 10 = 90 ms.

This means that spins within the echo have had 90 ms to dephase between their excitation pulse
and the regeneration of the echo. T2 weighting results.

Rephasing has been initiated by an RF pulse rather than a gradient so that more T2 information
is present. The rewinder gradient merely repositions the stimulated echo at a time when it can
be received.

Uses

SSFP sequences were used to acquire images that demonstrate true T2 weighting (Figure 5.32).
They were especially useful in the brain and joints with both 2D and 3D volumetric acquisitions.

Pulse sequences Chapter 5

177

Figure 5.32 Axial SSFP image through the brain.

FSE has now largely replaced this sequence as it produces better T2 weighting in short scan times.
However, the process of shifting the stimulated echo is used in sequences where rapid data acqui-
sition and long TEs are required. An example of this is in perfusion imaging (see Chapter 12).

Learning point: T2* vs true T2

It is important to understand the difference between the terms true T2 and T2*. This is best
demonstrated in imaging of the cervical spine. If the suspected pathology is a herniated disc,
then using a T2* gradient echo sequence such as coherent gradient echo is appropriate. The
disc will be demonstrated as low signal intensity disc bulge into a high signal intensity CSF-filled

Chapter 5 MRI in Practice

thecal sac and will produce a change in morphology (Figure 5.33). If, however, the pathology
is more subtle, for example a small MS plaque within the cord, then we need to use a true T2
weighted sequence where the contrast seen depends on differences between the T2 times of
the pathology and surrounding cord (Figure 5.34). In these circumstances it is better to use spin
echo type sequences such as CSE, FSE or SSFP that use TEs long enough to measure the T2

178 decay times of tissues present.

Parameters

To maintain the steady state:

• flip angle 30–45°
• TR 20–50 ms.

The actual TE affects the effective TE. The longer the actual TE, the lower the effective TE.
Actual TE should therefore be as short as possible.

• Average scan time – seconds for slice-by-slice acquisitions to several minutes for volumes.

Some manufacturers suggest decreasing the effectiveTE to reduce magnetic susceptibility,
and increasing the flip angle to create more transverse magnetization, which results in
higher SNR.

Advantages

• can be acquired in a volume and in 2D
• truer T2 weighting achieved than in conventional GE

Disadvantages

• susceptible to artefacts
• image quality can be poor
• loud gradient noise

Learning point: differentiating
common steady state sequences

As previously explained, the steady state produces two signals:

• a FID made up of transverse magnetization that has just been created
• a stimulated echo made up of the residual transverse magnetization component.

Pulse sequences Chapter 5

Coherent gradient echo, incoherent gradient echo and SSFP pulse sequences can be differenti- 179
ated according to whether they use one or both of these signals.

• Coherent gradient echo samples both the FID and the stimulated echo to produce either

T1 or T2* weighted images depending on the TE used (Figure 5.35).

• Incoherent pulse sequences samples the FID only to produces mainly T1 weighted images

(Figure 5.36).

• SSFP samples the stimulated echo only to produce images that are more T2 weighted

(Figure 5.37).

Balanced gradient echo

Mechanism

This sequence is a modification of the coherent gradient echo sequence that uses a bal-
anced gradient system to correct for phase errors in flowing blood and CSF, and an alternating
RF excitation scheme to enhance steady state effects. In addition, both the FID and the spin
echo are collected within a single readout. This results in images where fat and water produce
a higher signal, greater SNR and fewer flow artefacts than coherent gradient echo in shorter
scan times.

The balanced gradient system is shown in Figure 5.38. As the area of the gradient under
the line equals that above the line, moving spins accumulate a zero phase change as they
pass along the gradients. As a result, spins in blood and CSF are coherent and have a high signal
intensity. This gradient formation is the same as flow compensation or gradient moment
rephasing (see Chapter 6). In balanced gradient echo this gradient is applied in the slice and fre-
quency axes.

In addition, higher flip angles and shorter TRs are used than in coherent gradient echo, produc-
ing a higher SNR and shorter scan times. Normally this combination of flip angle and TR would
result in saturation and therefore enhanced T1 contrast. However, saturation is avoided by chang-
ing the phase of the excitation pulse every TR. This is achieved by selecting a flip angle of 90°, for
example, but in the first TR period only applying half of this, i.e. 45°. In successive TRs the full flip
angle is applied but with alternating polarity so that the resultant transverse magnetization is
created at a different phase every TR (i.e. 180° apart) (Figure 5.39). In this way, saturation is
avoided and fat and water, which have T1/T2 values approaching parity, return much higher signal
than tissues that do not. The resultant images display high SNR, good CNR between fat, water
and surrounding tissues, fewer flow voids and in very short scan times.

Uses

Balanced gradient echo was developed initially for imaging the heart and great vessels but is now
also used in spinal imaging, especially the cervical spine and internal auditory meatus as CSF flow
is reduced. It is also sometimes used in joint and abdominal imaging (Figures 5.40 and 5.41).

Chapter 5 MRI in Practice

180

Figure 5.33 Sagittal T2* weighted coherent gradient echo sequence through the cervical cord.
The prolapsed discs are well seen as they indent the thecal sac.

Pulse sequences Chapter 5

181

Figure 5.34 Sagittal T2 weighted FSE sequence through the cervical spine showing MS plaques
within the cord. It is possible that these may have been missed in a T2* weighted sequence
where the TE is not long enough to measure the T2 decay times of the pathology and the
surrounding cord.

Parameters

• large flip angle 90° (enhances SNR)
• short TR 10 ms (reduces scan time and flow artefact)
• long TE 15 ms (to enhance T2*)

182

Figure 5.35 Echo formation
in coherent gradient echo.

Figure 5.36 Echo formation
in incoherent gradient echo.

Figure 5.37 Echo formation
in SSFP.

Pulse sequences Chapter 5

Figure 5.38 Balanced 183
gradient system in
balanced gradient echo.

Figure 5.39 Maintenance of the steady
state in balanced gradient echo.

184

Figure 5.40 Axial
balanced gradient
echo image through
the abdomen.

Figure 5.41 Axial
balanced gradient
echo image through
the lumbar spine.